IZA DP No. 4410
FSU Immigrants in Canada: A Case of Positive Triple Selection?
Don DeVoretz Michele Battisti
September 2009
Forschungsinstitut zur Zukunft der Arbeit Institute for the Study of Labor
FSU Immigrants in Canada:
A Case of Positive Triple Selection?
Don DeVoretz
Simon Fraser University and IZA
Michele Battisti
Simon Fraser University
Discussion Paper No. 4410
September 2009
IZA
P.O. Box 7240
53072 Bonn Germany
Phone: +49-228-3894-0 Fax: +49-228-3894-180 E-mail: [email protected]
Any opinions expressed here are those of the author(s) and not those of IZA. Research published in this series may include views on policy, but the institute itself takes no institutional policy positions.
The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center and a place of communication between science, politics and business. IZA is an independent nonprofit organization supported by Deutsche Post Foundation. The center is associated with the University of Bonn and offers a stimulating research environment through its international network, workshops and conferences, data service, project support, research visits and doctoral program. IZA engages in (i) original and internationally competitive research in all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research results and concepts to the interested public.
IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.
IZA Discussion Paper No. 4410 September 2009
ABSTRACT
FSU Immigrants in Canada:
A Case of Positive Triple Selection?*
This paper investigates the economic performance of immigrants from the Former Soviet Union (FSU) countries in Canada. The contribution of this paper lies in its use of a natural experiment to detect possible differential labour market performances of Soviet immigrants prior to and after the collapse of the Soviet Union. In short, the collapse of the former Soviet Union allows an exogenous supply change in the number and type of FSU immigrants potentially destined to enter Canada. For this purpose, Census microlevel data from the 1986, 1991, 1996 and 2001 Canadian Census are utilized to estimate earnings and employment outcomes for pre- and post-FSU immigrants.
- JEL Classification:
- J61, F22
Keywords: immigration, integration
Corresponding author: Michele Battisti Department of Economics Simon Fraser University 8888 University Drive Burnaby, BC Canada
E-mail: [email protected]
*
We note with appreciation the copyediting services by M. Hayden of [email protected] and the financial support provided by the Research Authority, Ruppin Academic Centre, Israel.
3
Introduction
The post-1990 rise in immigration in general to Canada and from two disparate formerly closed systems—the Soviet Union and China—may have led to profound changes in the paradigm of economic integration into Canada’s labour force. Prior to the collapse of the Soviet Union, potential Soviet émigrés could not decide to move to Canada based on an open and easily transparent exit system. Thus, immigrants to Canada from the former Soviet Union (hereafter FSU) were largely designated by Canada as refugees and many came ill-equipped in terms of country-specific human capital to readily integrate into Canada’s labour market.
Figure 1
Source: LIDS (Landed Immigrants Data System) from IMDB Immigration Database
Figure 1 illustrates this point graphically. From 1980 through 1991 the distribution of FSU immigrants across entry gates was as follows: 58% refugees, 14.1% family class and 27.5% skilled class.1 By the year 2000, refugees made up only 13.9% of the entrants with 63.9% of FSU immigrants now appearing in the skilled group. In short, prior to 1992 FSU émigrés to Canada were only self-selected from the refugee portion of the potential pool of all FSU émigrés, while after 1992 FSU immigrants entered under a double selection system.
1 The skilled class potential entrant is assessed under a “points system” which yielded points for human capital attributes.
3
4
Figure 2
Source: LIDS (Landed Immigrants Data System) from IMDB Immigration Database
Figure 2 illustrates the uniqueness of the exogenous shock to FSU immigrant flow circa 1992-2001 when all immigrants to Canada had only a 5.3% drop in the proportion of refugees whilst the FSU immigrant share of refugees fell by 44.1%.
Thus, after the fall of the FSU (as with China circa 1995), immigrants who left the FSU were drawn from a larger pool of potential movers with a different set of observable human capital attributes. This important policy change in the FSU should ultimately reveal itself in differential labour force outcomes of FSU immigrants in Canada after 1991 if our thesis of positive selection holds. After 1991, the immigrants’ initial positive self-selection was combined with a second level of selection as these FSU émigrés were subjected to a “points assessment” system which favoured the admission of FSU immigrants with human capital.2 Thus, looking at the labour market performance of FSU immigrants entering before and after 1991 has the potential of shedding light on the effectiveness of Canada's selection process. The final or third selection process arises when the immigrant decides to ascend to Canadian citizenship or to remain a non-citizen.3
2
Pivnenko and DeVoretz (2003) note that a majority of Ukrainian immigrants to Canada came through non-economic entry gates prior to 1991.
3
Pivnenko and DeVoretz (2006) document the positive effect of citizenship status on the labour market outcomes of Ukrainian immigrants.
4
5
It should be noted that traditionally only a portion of “points-assessed” immigrants selfselect into citizenship, yet almost all refugees naturalize given their inability to return home. Given that refugees often feel compelled to naturalize, some of the economic premium owing to citizenship is often lost due to adverse selection; those FSU émigrés who arrived after 1991, however, were largely not refugees and should reveal a positive selection into citizenship since they were not compelled to naturalize.4 In fact, we argue that only those post-1991 FSU émigrés who acquired additional Canadian-specific human capital will tend to naturalize and reap the labour market rewards from acquiring this human capital.
In sum, the following thesis is offered in terms of the labour market integration of émigrés from the FSU into the Canadian context: prior to 1991, émigrés from the Soviet Union to Canada were singly selected by themselves and after 1991 the new cohort of FSU emigrants to Canada were often selected three times. This triple selection procedure in turn implies that a greater human capital stock will be embodied in this post-1991 cohort and would lead to more rapid integration into Canada’s labour market in the absence of discrimination or other forms of labour market failure. It is the purpose of this study to test this thesis in the context of a “gap analysis” in terms of income and employment.
The traditional immigrant earnings literature owing to Chiswick (1978) argues that upon entry, immigrants suffer an earnings deficit due to the absence of specific and general (language, knowledge of institutions) human capital attributes. It was inferred by Chiswick from census data that over time—generally 8 to 12 years—immigrants overcame these human capital deficits by investing in themselves and their earnings subsequently “caught-up” to and then perhaps surpassed their Canadian-born colleagues.
4 Pivnenko and DeVoretz (2006) verify this empirically for all Canadian refugees circa 2006 in Canada.
5
6
Figure 3: Idealised Age-Earnings Profile
Figure 3 depicts the “gap” hypothesis from both optimistic and pessimistic viewpoints. Given our thesis of “triple selection” we would expect that Figure 1 would apply to highly skilled FSU émigrés to Canada since increased observable human capital attributes owing to triple selection should hasten the diminution in the earnings gap and may lead to its complete evaporation at X. Beyond X, in the optimistic case the immigrant now can overachieve with respect to their Canadian-born cohort’s earnings performance. However, if there exist “unobservable” factors which intervene in this process of labour market integration, the case of underachieving may arise. These “unobservables” include inhibitions on the immigrant’s desire to self-select into the labour market, employer discrimination of the immigrant’s human capital characteristics (i.e. foreign education) and discounting their foreign labour market experience. It is this “gap” framework as depicted in Figure 1 which will inform our labour market integration analysis given the triple selection thesis outlined above. It should be noted that FSU immigrants may achieve over- or under-achievement status depending on the presence or absence of unobservable factors.
Literature Review
Canadian literature on the economics of immigration provides an extensive empirical immigrant labour market integration (Reitz, 2001). The literature has largely focused on Canadian immigrant earnings’ performance in general but a series of in-depth studies based on the immigrants’ country of origin have recently appeared. Two major findings from the general Canadian immigrant earnings experience appear to date. First, an age
6
7earnings profile analysis based on a human capital model forms the underlying analytical framework for immigrant labour market performance across entry groups and over time in Canada. Secondly, and this is more germane to this study, more recent immigrant entry cohorts have failed to “catch-up” to their Canadian-born cohorts while older vintages of immigrants have overachieved.
Figure 4: Age-earning Profiles
35000 30000 25000
CB
20000
BritIm_C
15000
BritIm_NC
10000
ChinIm_C
5000
ChinIm_NC
0
- 25
- 35
- 45
Age
- 55
- 65
Legend: CB: Canadian Born; BritIm_C: British Immigrants Canadian citizens; BritIm_NC: British Immihrants non-citizens of Canada; ChinIm_C: Chinese Immigrants Canadian citizens; ChinIm_NC: Chinese Immigrants non-citizens of Canada (ChinIm_NC)
Source: Census of Canada, 1996
Figure 4 presents two interesting empirically-based variants of the gap model illustrated in Figure 3 in the Canadian context.5 First, British immigrants circa 1996 could be termed ‘overachievers’ since they never suffered an earnings entry penalty and upon gaining citizenship outperformed their Canadian-born cohort every year over their life cycle. This overachieving phenomenon is repeated by immigrants from the United States and several other western European immigrant groups in Canada (Pivnenko and DeVoretz, 2003).
However, there exists a large dissenting literature which argues that the post-1990 wave of Canadian immigrants have not performed as described above. These observers report that each successive wave of post-1990 immigrants had a larger earnings entry penalty
5
Since these findings were derived from a series of pooled Canadian Censuses care in this interpretation must be made since aging, cohort and time in Canada effects are difficult to disentangle.
7
8and rarely overcame this increased penalty with time in Canada (Li, 2003). In addition, work on discriminatory behaviour in the Canadian immigrant labour market argued that institutional barriers prevented credential recognition (Ferrer and Riddell, 2002) and when coupled with overt discrimination (Pendakur and Pendakur 1998) prevented the post-1990 wave of Canadian immigrants from successfully integrating into the Canadian labour market. In Figure 4 these ‘underachievers’ appear in the form of the most recent wave of highly-educated Chinese immigrants in Canada. In fact, as reported in Figure 4, Chinese immigrants with or without citizenship status never ‘catch-up’ and if they do not ascend to citizenship they will still live below the poverty line for their entire lifetime.
The aim of this condensed literature review is to choose which strain of the above-noted gap analysis –over or underachieving—applies in the context of immigrants from the FSU to Canada. We currently have two econometric studies which address the economic performances of select groups of Canadian immigrants from part of the FSU. Pivnenko and DeVoretz (2003) investigated the economic performance of recent Ukrainian immigrants to Canada and the United States with available census data. Their underlying approach was to test for ethnicity, foreign birth status and destination effects on the economic performance of a pooled set of pre- and post-1990 FSU Ukrainian immigrants in Canada and the United States. In particular, they tested for the existence of earnings overachieving in the context of Ukrainian immigrants in North America. An important sub-hypothesis is also addressed when they speculate that Ukrainian immigrants overachieve because they enjoy a ‘sheepskin effect’ which raises Ukrainian immigrant earnings relative to other immigrant graduates because employers may value Ukrainian degrees more. Finally, Pivnenko and DeVoretz conducted a comparative analysis of Ukrainian immigrant earnings in the United States versus Canada to detect if Canada’s highly selective immigration policy encouraged more productive immigrants to enter Canada from the Ukraine.
Their reported results indicate that recent Ukrainian immigrants to Canada are indeed a select group. For the period 1991-2001, Ukrainian immigrants to Canada arrived with higher educational attainment, a greater propensity to speak English at home and contained the largest percentage of professionals for any immigrant cohort over the 1991- 2001 period. These human capital attributes led to above-average earnings performance for Ukrainian immigrants which in turn was explained by their occupational distribution (largely professionals), numbers of weeks worked, and a substantial ‘Sheepskin Effect’.6
The result of this robust earnings function is that Ukrainian immigrants in Canada outperformed the earnings of all other Canadian immigrants and “caught-up” and surpassed their Canadian-born cohort at age 36 as depicted in Figure 5.
6
In fact, Pivnenko and DeVoretz (2003) report that this earnings effect derived from completing a university degree was the greatest for Ukrainian immigrants relative to all other Canadian immigrants.
8
9
Finally, Pivnenko and DeVoretz (2003, p.13) conclude from their study that:
“For Ukrainian immigrants, the assimilation process starts at a higher income level that exceeds the income earned by non-Ukrainians with the difference growing over time. The greater intercept reflects the more favorable entry effect for the Ukrainian immigrants. This positive earnings premium implies that…..the quality of the earnings enhancing characteristics Ukrainians have acquired…is relatively higher than for the rest of the immigrant population.”
In other words, Ukrainian immigrants in general were overachievers. They further report that Ukrainian immigrants to the United States do even better because they were endowed with greater human capital than Ukrainian immigrants resident in Canada.
However, we must be cautious not to draw hasty conclusions from this Ukrainian study for the FSU immigrant experience in general. First, the Pivnenko-DeVoretz sample is restricted to Ukrainians only and includes both pre- and post-FSU populations of all skill types. It is possible that the subject of this study, namely post- FSU arrivals, will exhibit a pattern of underachieving that appears in Figure 3.
Dean and DeVoretz (2000) with a similar analysis explored the “gap” thesis for all Jews living in Canada circa 1996. Again, this population does not match the former FSU immigrant stock which is the focus of this study but does include many former FSU immigrant arrivals. Dean and DeVoretz (2000) ask whether ethnicity (i.e. Jewish or nonJewish) is related to the economic performance of immigrants in Canada. Their underlying argument is that income-enhancing non-cultural characteristics (e.g. education) are correlated with cultural characteristics. Using Canadian census data, their study group overlaps with the immigrant sample considered in this study, namely those
9
10 immigrants from the FSU.7 They isolated two Jewish sub-groups: Jewish Canadian-born and Jewish immigrants and estimated earnings functions for these two groups as well as their non-Jewish counterparts. They reported that the stock of human capital characteristics which were normally correlated with higher income (age, education, and English language skills) exceeded all other immigrant groups to such an extent that any earnings entry penalty owing to immigrant status was overcome by Jewish immigrants by virtue of other income correlates. For example, almost 100% of Jewish immigrants reported speaking English at home while only 69% of all immigrants reported a similar capability. In addition, Jewish immigrants are older and more likely to be married than non-Jewish immigrants. However, one glaring inconsistency occurs when they observe that Jewish immigrants have less education than their Canadian-born Jewish counterparts. Nonetheless, Jewish immigrants are highly concentrated in the professions in Canada. In terms of gender, it is reported that circa 1990 Jewish immigrant women opted out of the Canadian labour market and, when in the labour force, were more likely to work for wages and salaries.
Their regression analysis of the earnings model allowed Dean and DeVoretz (2000) to conclude that both the substantial human capital endowments of Jews born in Canada and the differential rewards paid to these educational endowments allowed Jewish-Canadian immigrants resident in Canada circa 1990 to outperform other immigrants and avoid an earnings penalty upon entry into Canada.
At this point in the review, we conclude that two econometric studies which partially cover our immigrant group of interest (FSU) indicate that prior to 1995 Jewish and Ukrainian immigrants to Canada were exceptional groups. They earned more than their other foreign-born cohorts in Canada due to either greater human capital endowments or a better recognition of their credentials, or both.
Data
Data source
The data we use in this paper are drawn from the individual Public Use Micro Files (PUMF) from the Canadian Census of Population for the years 1986, 1991, 1996 and 2001. These datasets contain information on a representative sample of people living in Canada in the years 1985, 1990, 1995 and 2000 respectively. The total sample sizes of the PUMFs are 500,434 for the 1986 Census, 809,654 for the 1991 Census, 792,448 for the 1996 Census, and 801,055 for the 2001 Census.
The choice to work with census data has a number of inherent disadvantages. Census data is not panel data, so we cannot follow the same people over time, and thus in our econometric estimation we cannot control for unobservables that happen to be correlated with the variable that identifies FSU as the origin region. In addition, stacking four
7 In fact, in the 1991 Canadian 2% PUST a cross tabulation of the Jewish sample indicates that over 60% of recent Jewish immigrants originated from the FSU.
10
11 different Censuses together introduces a number of possible sources of bias which makes the data cleaning process for this paper particularly complex.
Nonetheless, the individual-level Census data seem to be the best choice to analyze our research question. The most important single reason for this conclusion is that it is the only dataset that leaves us with a large number of observations for FSU immigrants that entered Canada before and after 1991. 8
Construction of our dataset
Unfortunately, a few variables of interest for our study are coded differently across the different censuses. In other words, the coding system adopted for the construction of most of the categorical variables vary across different censuses, so that it is not possible to simply stack the data. The information contained in our data set is ultimately equivalent since we will only include variables for which this is the case. In the very few cases in which the recoding procedure does have an impact on the informational content, these differences will be explicitly mentioned.
The procedure to create a consistent dataset used the 2001 Census coding as a starting point, and subsequently modified the coding of all other Censuses to make it comparable to the 2001 Census. Variables identifying birthplace and those identifying year of immigration do not have a perfectly overlapping coding across census years, however we did recode to avoid any bias.9
PUMF files for each census report variables are expressed in Canadian Dollars (CAD) as reported by the respondents for one year prior to the release of the relevant census, therefore all monetary variables employed in our estimates (wages and salaries, selfemployment income, total income, government transfers) must be adjusted for inflation. For the reported statistics in our summary tables and our regressions, all monetized variables are expressed in dollars for the year 2000 leaving the monetized values for 2001 unchanged.10
Data selection
For the summary statistics and for our wage equations, we restricted our sample to individuals of working age (i.e. aged 20-65) and excluded individuals for whom the primary source of reported income is self-employment income.
8
In particular, the publicly available version of the IMDB dataset does not offer a sufficiently large sample size and, even more crucially, does not include immigrants who arrived in Canada before 1981. 9 Codes for the construction of the merged dataset are available from the authors